This analysis reports the first evidence of the parental-offspring BMI associations across three generations where mothers were found to have a stronger BMI association with their progeny. We also compared the cross-generation pattern for BMI with the cross generation pattern for height across three generations. While only the maternal line indicates a cross-generation BMI transmission, both maternal and paternal lines appear to contribute to offspring height. These results confirm previous findings [6
] where a considerable amount of the variance of BMI is related to genetic and shared environmental factors but there remains a proportion that is directly attributable to the maternal line of the family. Our results show for the first time that maternal influence is present over three generations. BMI in early childhood is, arguably, therefore influenced to some degree by maternal specific effects, possibly due to intrauterine exposures, which is in keeping with evidence from animal studies [4
To our knowledge only two previous human studies have reported overweight and obesity in three generations. In the Belgian-Luxembourg study children's BMI was related to BMI of both parents and obesity measurements in their grandparents [15
]. However the measure of obesity in grandparents used in the study was an alternative measure to height and weight and may not accurately reflect the strength of the association. Davis et al. in their study of 2591 US children grouped parental data and grandparental data to generate composite measures of BMI [16
] which does not distinguish between the maternal and paternal lines. Neither does it allow simultaneous estimation of the effect of explanatory variables as described here.
Several longitudinal studies have examined BMI across two generations. The British 1958 birth cohort demonstrated how parental BMI during childhood and adulthood was clearly associated with offspring BMI [29
]. However, the relative contributions of the parents could not be examined as BMI data was only available for the cohort member and not their partner. In addition BMI zscores were used which has inherent disadvantages. In our mixed model analysis, this problem is solved simply by the inclusion of a group term in the models and by modelling different variances in groups via the correlation structure.
Davey Smith and colleagues examined the BMI relationship between parent and child and found no difference in the relative contributions of maternal or paternal BMI to offspring BMI [5
]. Indeed, if the heritability estimates of mothers' BMI with both of her parents in the present study were reported alone, this would indicate the paternal and maternal effects contribute equally to offspring BMI. Furthermore, the maternal effects are stronger between the older generation compared with the younger. The differences in findings between other studies [5
] and those presented here may be explained by the specific ages of offspring under study. Between five and seven years children experience varying changes in growth, defined by the critical period of adiposity rebound [30
]. We speculate that the age of onset of this growth period may be determined by maternal effects with a certain window of influence from birth to adiposity rebound. Therefore our findings of a maternal specific heritability of BMI may be particularly evident in early childhood.
Findings from the Early Bird 43 study have demonstrated a gender specific association between mothers and daughters and fathers and sons from the ages of 5 to 8 years [6
]. We did not find evidence of a gender effect on the child but the findings from the Early Bird study indicate the possible behavioural role models of parents which appear to strengthen with age [6
]. The association between mother and offspring may not be evident in the Early Bird 43 study as firstly the mother's BMI was recorded in cross-section when the child was aged 5 and secondly older children will have longer exposure to environmental factors shared with their parents and therefore demonstrate stronger associations with their parents than younger children [15
Several studies examine the relationship of mother-offspring BMI, birth weight and other anthropometric measures [2
]. Fewer studies include fathers which limits the interpretation of their results; inclusion of fathers provides a greater understanding of the genetic effects as the father-child relationship has less opportunity for confounding with environmental effects. The heritability estimates of height in the present analysis are clearly larger than for BMI indicating a strong genetic contribution to height. Moreover, the use of height in this study enabled us to demonstrate that the dataset is large enough to differentiate the patterns of BMI and height between generations.
Our study supports previous evidence of the social patterning of height but not of BMI which is in contrast to previous studies [33
] and furthermore markers of diet and exercise were no longer significant. It is possible that these indicators may be too crude to elucidate clear lifestyle differences, however most measures of diet and physical activity will be influenced by socioeconomic status. What is clear from our findings is that despite the inclusion of socioeconomic indicators there remains a fundamental link between the BMI of mother and child that is not explained by shared environmental factors alone. Furthermore, although certain risk factors were not found to be significant in these models the strong family component would suggest that modifiable risk factors, such as physical activity, may be more likely to be successful if they are targeted at the family level rather than at an individual level.
Strengths and limitations
The advantages of this study are the prospective design with BMI information collected prior to pregnancy and in early childhood, in addition to data from three generations of one family. To our knowledge this is the first prospective study to show intergenerational associations between BMI in the youngest generation and BMI in the previous two generations, with individual measurements for all four grandparents.
Ideally, the strength of our findings could be improved by a higher response rate and larger cohort numbers. Nevertheless, in a previous analysis we did not find any differences in pre-pregnancy BMI of the mothers who responded to the follow-up compared to non-responders. Furthermore, an investigation into the pattern of missing data did not demonstrate any systematic variability. A potential limitation of this study was that self-reported adult height and weight measures were used and, while such measures are acceptable, may lead to underreporting bias particularly in women [35
]. This means in fact we may be under-estimating the strength of the effects found since the actual BMI of mothers and grandmothers is likely to be higher.
In epidemiological terms this is a relatively small cohort however there are no comparable cohorts with seven members of the one family. Limited families within the study have complete information on all family members but the mixed model maximises the use of the available data. Furthermore, despite the smaller sample numbers of fathers and grandparents the analyses have sufficient power to demonstrate different familial relationships for BMI and height.